The Effects of Stock Lending on Security Prices: An Experiment

advertisement
The Effects of Stock Lending on Security Prices:
An Experiment
Steven N. Kaplan,* Tobias J. Moskowitz* and Berk Sensoy**
*University of Chicago and NBER and **Ohio State University
1
© 2010
Motivation: What is the impact of short selling?
!
Does shorting make prices more efficient by reducing overpricing?
– Miller (1977) predicts differences of opinion and short sales
constraints can lead to overpricing
– Diamond and Verrecchia (1987) argue traders will adjust to short
sale constraints so that there is no overpricing
!
Does shorting make prices less informative and destabilize markets?
– Moving prices further away from fundamentals [Lamont (2004)]
– Help reduce excess volatility [Hong and Stein (2003), Abreu and
Brunnermeier (2001), Allen and Gale (1991)]
2
© 2010
Effect of short selling is an empirical question
!
Attempts to measure the empirical effects of short selling from both the
demand and supply side.
– Holding supply constraints fixed, variation in shorting demand can
measure the extent of overpricing or excess volatility
– Holding demand fixed, variation in shorting supply can measure
degree to which constraints matter
!
Empirical efforts hampered by difficulty of identifying pure demand or
supply shocks.
– E.g., short interest, rebate rates, loan fees
– Regulation shocks (2008 Emergency Order, SHO, shorting bans
around the world)
3
© 2010
Previous Empirical Work
!
Shorting demand:
– Use direct measures of shorting costs (rebate rate or spread
between the rebate and interest rates) [D’Avolio (2002), Geczy,
Musto, and Reed (2002), Ofek, Richardson, and Whitelaw (2004),
and Jones and Lamont (2002)]
– Short interest [Desai et al. (2002) for a summary]
!
Shorting supply:
– Ofek and Richardson (2003) consider lockup expirations
– Chang, Cheng, and Yu (2007), Bris (2008), Diether, Lee, and
Werner (2009), Beber and Pagano (2010) on regulation changes
!
Cohen, Diether, and Malloy (2008) look at both.
!
Mixed results---hard to separate supply
and demand.
4
© 2010
How Does Short Selling Work?
(Cohen et al. (2004))
!
!
!
!
!
Investor who wants to short must find a lender. Sells shares short.
Then:
– Short seller must deliver shares to buyer by day T+3.
– Short seller must maintain margin of >= 30% of short position.
Proceeds of short sale deposited with lender.
– Lender requires 102% of value of shares as collateral.
» Fluctuates with value of shares.
Lender invests collateral and earns interest.
– Interest is reinvestment rate.
Lender also pays borrower a rebate rate which is <= reinvestment rate.
– Rebate rate can be negative.
Lender’s profit = Reinvestment rate - Rebate rate = Loan Spread / Fee.
– Stocks in highest demand (special) can have negative rebate rates.
5
© 2010
!
Recall:
– Lender can recall the shares at any time.
– Borrower must return shares in 7 business days.
6
© 2010
What do we do?
!
Working with an anonymous money manager (greater than $15 B in
assets) we randomly:
– make available for lending 2/3 of Manager’s stocks and withhold a
characteristic-matched other 1/3 of Manager’s stocks
– focus on high loan fee stocks (loan fee > 25bps, with mean > 4%)
!
Provides exogenous shock to supply of lendable shares
– Supply shift not driven by changes in “The Manager’s” marginal
lending cost
– No plausible change in shorting demand
– Focus on stocks with high loan fees, consistent with theory and
empirical work suggesting where effects mostly reside
[Duffie, 1996; Duffie et al. 2002; Kolasinski et al. 2010].
7
© 2010
Experimental Design and Sample
!
The Manager
– Invests in mid and small-cap equities, inside and outside U.S.
– Had not lent out stocks out of concern that doing so would:
» lower the stocks’ prices
» increase their volatility
!
Motivation for experiment
– Consider fees / benefits from lending shares
vs.
– Adverse effects / costs
– Measure magnitude of net benefit/cost
8
© 2010
Experiment Phase 1
!
!
!
Shares available for lending on September 5, 2008.
– Selected sample based on stockholdings as of June 30, 2008
– Manager owned 523 stocks worth in excess of $15 billion
Stocks divided into two groups:
1. High expected loan demand group included stocks projected to
have loan fee of at least 10 basis points(138 “Revenue stocks”)
2. Remaining group of 385 stocks we refer to as low demand or "nonrevenue" stocks.
Within each group, randomly selected to lend out 2/3 and withhold 1/3
– One exception - lent three stocks in revenue stock group with the
highest expected revenue to reduce opportunity cost
– Results are the same if we exclude these three stocks
9
© 2010
!
Figure 1 - distribution of firm characteristics across the revenue stocks
available to lend out and revenue stocks withheld
– No significant differences between available and withheld groups
!
Figure 2 – (same for) non-revenue stocks
– No significant differences between available and withheld groups
!
Randomization seems to work---observables no different across the
treatment and control groups
10
© 2010
Additional Restrictions
!
!
!
!
!
!
Shares traded on U.S. exchanges
Shares that were in high demand
– revenue stocks with gross loan fees of at least 25 basis points
Loan size restricted to the lesser of:
– three times the average daily trading volume (in past 30 days)
– 5% of outstanding shares of issuer
Additional restrictions yield 32 available and 20 withheld stocks.
Concern: puts a limit on increase in supply from experiment
Response:
– these stocks are those with greatest shorting demand, so supply
shocks should matter most for these stocks
– Most variation comes from ownership and NOT these restrictions
11
– Cross-sectional results
© 2010
!
Manager supply shock was likely only supply shock by any mutual fund
during this period.
– Rizova (2010) examines securities lending participation and
revenue for most mutual funds (based on mutual fund annual
reports on SEC Edgar website)
– Only one mutual fund changed its participation and share supply to
the loan market over the first phase
» Our Manager
!
While we cannot rule out change in participation of other institutions,
seems likely our experiment was only major supply shock.
12
© 2010
Experiment Phase 2
!
We repeated the experiment from June 5 to September 30, 2009.
– Same restrictions applied (except 25 basis points throughout)
– Same randomization process
– End up with 19 available stocks and 9 withheld stocks
!
On October 1, all shares were made available for lending with a limit of
$500 million
13
© 2010
This paper
!"#$%#&'(""')*+
!"#$%#&',-../0
!"#$%#&'$"12#$
314-#5'/42#"$'46',7465'%#5"6",5')8+
14
© 2010
Shorting constraints and share prices
!"#$%&'()$*
+),-.(**/*0"
&&&&&&&&&&&&&&&&&&1#2*3,
4-")#0-3
56'*$"-")#0,
1#2*3,
!7#(")0.&$#0,"(-)0",
15
© 2010
Sample Description – Table I
!
Compare available stocks to withheld stocks on observable dimensions
!
No differences across a variety of characteristics
!
For first phase, only significant differences are higher mean (but not
median) institutional ownership and marginally higher short interest
!
For second phase, only significant difference is higher expected loan fee
for available stocks
!
Consistent with random variation
!
But, if anything, differences will likely bias towards finding an effect
16
© 2010
Table I: Randomization
17
© 2010
Table I (continued)
18
© 2010
The Lending Experiment and Summary Statistics
First phase:
! Lending period began September 5, 2008.
– At peak, on September 17, over $700 million of securities lent out.
!
September 18, 2008, Manager asked lending agent to call loans back.
– Last shares returned on October 3, 2008.
!
We examine effects on stocks lent out vs. those withheld during:
– "lending period" (September 5 to 17). Max. 9 trading days.
– "recall period" (September 18 to October 3). Max. 12 trading days.
!
“Pre-period” is August 1-Sept. 4, 2008.
19
© 2010
The Lending Experiment and Summary Statistics
Second phase:
!
Lending period began June 5, 2009.
– Up to $200 M lent out through August 7.
– After August 7, up to $350 M lent out (and limit not reached) through
September 30.
!
Withheld stocks were made available on Oct. 1.
!
“Lending period” is June 5 – Sept. 30.
!
“Pre-period” is May 1 – June 4, 2009.
20
© 2010
Table II: Summary Statistics
21
© 2010
Does Experiment Provide a Sizeable Shock to Supply?
!
Potential loan from experiment is:
» For first phase, mean of 2.3 times average daily volume, 3.7% of
institutional ownership, 18.3% of short interest
» For second phase, mean of 2.3 times average daily volume, 6.9%
of institutional ownership, 36.8% of short interest
!
Increase in supply should have largest effect for stocks in our sample -high demand and illiquid [Kolansinki, Reed, and Ringgenberg (2010) ]
!
And, at least for first phase, we know experiment was likely the only
shock (from Rizova’s (2010) data)
22
© 2010
Does Experiment Provide a Sizeable Shock to Supply?
!
Supplemental data from Data Explorers on loan fees and quantities
» Available stocks hit by our supply shock
» Withheld stocks not shocked
!
For both lending and recall periods (and combined), strong positive
correlation between number of shares we know were on loan and
number of shares on loan reported by Data Explorers
!
Strong positive correlation between number of new shares we lent on a
given day and the corresponding number from Data Explorers
!
Fees the Manager received highly correlated with Data Explorers
average fees, but no perfectly so (0.69)
"Segmented market
23
© 2010
Does Experiment Provide a Sizeable Shock to Supply?
!
What happens to market loan fees for the available (treatment) stocks
versus withheld (control) stocks?
– If supply shock big enough, loan fees should decline significantly
– Size of decline indicates importance of supply change
– Table III
» Measure loan fee changes for available vs. withheld stocks
» Difference-in-differences approach comparing pre-lending fees
and fees during the lending program
» Use multiple measures of fees from multiple sources
• Data Explorers
• Lending agent
• Actual fees received
» Examine cross-section of size of loan
24
© 2010
Table III: Difference-in-Differences for Loan Fees
Pa ne l A: Loa n Fe e Diffe re nce s-in-Diffe re nce s (%)
First pha se
Available - Withheld
Available - Withheld
mean
Lending period fee
! pre-lending period fee
(from Data Explorers)
-0.09
Lending period fee (from Data Explorers) –
-0.70
expected fee (from lending agent)
[1.28]
Actual lending fee – expected fee (from
lending agent)
-3.45**
Actual lending fee – pre-lending period fee
-2.84***
Se cond pha se
Available - Withheld
Available - Withheld
median
mean
0.05
-2.29*
[0.45]
-0.56**
[1.13]
0.02
-2.27*
-0.48
[1.14]
-0.50
-3.01**
-0.94**
[1.30]
[1.46]
(from Data Explorers)
median
-1.82***
-3.03**
[0.66]
-1.19**
[1.30]
Pa ne l B: Re gre ssions of Cha nge s in Loa n Fe e on Loa n Size
First pha se
Available stocks only
Actual lending fee
Dependent variable =
Actual loan/potential loan (% )
! pre-lending fee
Actual lending fee
! pre-lending fee
-0.34***
-0.35***
-0.18**
-0.17**
[0.12]
[0.09]
[0.07]
[0.06]
Actual loan/shares outstanding (% )
N
R -square
Se cond pha se
Available stocks only
All stocks
All stocks
-3.55
[3.37]
32
0.28
32
0.03
-6.91**
[2.94]
52
250.42
52
0.15
-6.09***
[1.53]
19
0.50
19
0.78
-5.83***
[1.41]
28
0.54
28
© 2010
0.79
Table IV: Return Differences
!
Portfolio approach to avoid cross-correlation problem
– Three sets of weighting schemes:
» equal weight
» value weight
» weight by expected loan fee before the lending experiment
!
If loan supply has pricing effect (at amounts being lent), then expect
» decrease in returns of available relative to withheld stocks
during lending period (when supply is exogenously increased)
» reversal or increase in returns during recall period (when
increase in supply is taken away)
26
© 2010
Table IV: Return Differences
!
Available
Pre-period
Withheld Difference
Lending period
Available
Withheld Difference
Recall period
Available
Withheld Difference
Panel A: First Phase
Equal-weight
0.16
[0.47]
0.15
[0.47]
0.01
[0.26]
(0.98)
-0.26
[0.86]
-0.72
[1.05]
0.47
[0.37]
(0.24)
-0.69
[1.50]
-0.27
[1.49]
-0.42
[0.50]
(0.42)
Value-weight
0.14
[0.48]
0.01
[0.40]
0.13
[0.29]
(0.67)
-0.19
[0.86]
0.09
[0.94]
-0.28
[0.47]
(0.57)
-0.70
[1.47]
0.01
[1.46]
-0.71
[0.77]
(0.38)
Expected loan fee-weight
0.23
[0.52]
-0.09
[0.60]
0.32
[0.41]
(0.44)
0.31
[0.84]
-0.80
[0.93]
1.11***
[0.28]
(0.00)
-0.85
[1.59]
-0.55
[1.58]
-0.29
[0.49]
(0.56)
Panel B: Second Phase
Equal-weight
0.67
[0.66]
0.44
[0.53]
0.23
[0.31]
(0.46)
0.36
[0.23]
0.23
[0.20]
0.13
[0.13]
(0.35)
Value-weight
0.53
[0.56]
0.67
[0.51]
-0.14
[0.35]
(0.69)
0.34
[0.20]
0.13
[0.19]
0.21*
[0.13]
(0.09)
Expected loan fee-weight
1.59
[1.02]
0.27
[0.61]
1.32*
[0.75]
(0.09)
0.35
[0.31]
0.22
[0.21]
0.13
[0.24]
(0.59)
27
© 2010
Power of the Tests
!
Each phase individually has only moderate statistical power
!
Since two phases are independent, joint power is quite high
– Consider Miller’s (1977) overvaluation hypothesis that returns during
the lending period are negative (a one-tailed test)
» equal-weighted results, the joint probability observe at least
• 47 basis points (p-value = 0.12) in first phase (trial)
• 13 basis points (p-value = 0.175) in second phase (trial)
» relative to the null that returns in each phase are less than or
equal to zero = 2.1%
!
Alternatively, can ask what level of returns can we reject at
conventional levels of significance taking into account both phases?
– Rejection regions that combine both phases---Table V
28
© 2010
!
Table V: Power of the Tests
Rejection Cutoff Values for Return Differences (in %)
First phase
Significance
level
Lending
(lower bound)
Recall
(upper bound)
Second phase
Lending
(lower bound)
Combined phases
Lending
(lower bound)
Equal-weighted
10%
5%
1%
-0.04
-0.21
-0.60
0.26
0.48
0.94
-0.05
-0.10
-0.19
0.01
-0.03
-0.12
Value-weighted
10%
5%
1%
-0.94
-1.16
-1.65
0.34
0.67
1.38
0.05
0.00
-0.08
0.01
-0.03
-0.12
Expected loan fee-weighted
10%
5%
1%
0.72
0.59
0.30
0.38
0.59
1.04
-0.18
-0.27
-0.43
0.31
0.25
0.12
29
© 2010
Volatility, Skewness, and Bid-Ask Spreads
!
Differences-in-differences between stocks made available vs. those
withheld from
» pre-lending to lending period
» lending period to recall period
– Cross-sectional average of volatilities and skewness from daily
returns over each period
– Time-series average of daily bid-ask spreads of each stock (as %
price) over the specified period
!
If short sale constraints are important for price discovery:
– Expect decrease in volatility, skewness, and bid-ask spread over the
lending period
– Expect reversal over the recall period
If short sales are destabilizing, expect the opposite pattern
!
30
© 2010
Table VI:
Changes in Volatility, Skewness, and Bid-Ask Spread
Differences-in-Differences Between Available and Withheld Stocks from Pre-, Lending, and Recall Periods
Volatility differences
First phase
Lending Recall pre-period
lending
Second phase
Lending pre-period
Skewness differences
First phase
Lending Recall pre-period
lending
Second phase
Lending pre-period
Bid-ask spread differences
First phase
Lending Recall pre-period
lending
Second phase
Lending pre-period
Equal-weight
-0.82
(0.14)
0.50
(0.59)
-0.59
(0.38)
0.09
(0.80)
-0.06
(0.78)
-0.46
(0.43)
-0.04
(0.36)
-0.68**
(0.04)
-0.04
(0.31)
Value-weight
-0.54
(0.50)
-0.31
(0.79)
-0.48
(0.39)
0.08
(0.86)
0.07
(0.82)
-0.56
(0.28)
0.00
(0.88)
-0.33**
(0.03)
-0.04
(0.18)
Expected loan fee-weight
-0.07
(0.94)
0.01
(1.00)
-1.15
(0.49)
-0.13
(0.85)
0.48
(0.29)
-0.68
(0.19)
-0.05
(0.63)
-0.96
(0.13)
-0.08
(0.30)
31
© 2010
Post-October 1, 2009 for 2nd Phase
!
After October 1, 2009, manager lifted lending restrictions on withheld
stocks.
!
No significant effects on returns, volatilities, skewness, spreads over
ensuing 3 months.
!
Almost a “third” independent phase for a positive supply shock (this
time on the formerly withheld stocks)
32
© 2010
Cross-sectional Results
!
Average results not consistent with overvaluation hypothesis from short
sale restrictions or destabilizing effects from stock lending
!
Possible that average results mask important cross-sectional effects
!
Interact whole host of stock and Manager holdings characteristics with
a dummy variable = 1 if stock is available
» Are there greater effects for stocks in which demand is high
relative to supply?
» Are there greater effects for larger supply shocks?
33
© 2010
Table VII: XS-Regression Lending Period
First Phase
Dependent variable:
Second Phase
Return
Return
Panel A: Lending period
Short interest
0.05
[0.37]
-0.08**
[0.04]
-0.03
[0.15]
-0.01
[0.12]
-0.01
[0.63]
-0.03
[0.64]
-0.06
[0.32]
0
[0.85]
Potential loan to inst. own. (%)
0.15
[0.52]
0.14
[0.27]
-0.24***
[0.00]
-0.01
[0.43]
-0.04
[0.33]
-0.18
[0.40]
-0.17
[0.17]
-0.01*
[0.08]
Expected loan fee
-0.04
[0.81]
-0.11
[0.17]
0.04
[0.45]
0.01
[0.61]
0.06
[0.74]
-0.88
[0.18]
0.21
[0.79]
-0.01
[0.76]
Short interest * Available
-0.02
[0.77]
0.03
[0.48]
0.04
[0.17]
0.01
[0.12]
0.01
[0.49]
0
[0.98]
0.06
[0.39]
0
[0.50]
Potential loan to inst. own. (%) * Available
-0.37
[0.22]
0.05
[0.79]
0.24
[0.14]
0.01
[0.43]
0.04
[0.31]
0.2
[0.35]
0.22*
[0.07]
0.01
[0.21]
Expected loan fee * Available
0.06
[0.72]
0.16*
[0.07]
-0.06
[0.25]
0
[0.71]
-0.06
[0.73]
0.74
[0.26]
-0.27
[0.74]
0
[0.95]
Available
1.81
[0.35]
-1.81
[0.19]
-1.32
[0.17]
-0.22
[0.14]
-0.18
[0.66]
-1.83
[0.33]
-2.09
[0.24]
-0.02
[0.78]
N
R-squared
52
0.14
52
0.3
52
0.18
52
0.17
28
0.12
28
0.33
28
0.2
28
0.14
34
© 2010
Table VII: XS-Regression Recall Period
Dependent variable:
Return
Panel C: Recall period
Short interest
-0.04
[0.12]
0.07
[0.21]
0.00
[0.81]
-0.04**
[0.02]
Potential loan to inst. own. (%)
-0.13
[0.32]
-0.33
[0.15]
0.01
[0.83]
-0.1
[0.17]
Expected loan fee
-0.01
[0.83]
0.15
[0.25]
-0.05*
[0.05]
0.04
[0.18]
Short interest * Available
0.01
[0.87]
-0.01
[0.88]
0
[1.00]
0.03
[0.11]
Potential loan to inst. own. (%) * Available
0.01
[0.95]
0.46
[0.17]
-0.06
[0.54]
0.07
[0.64]
Expected loan fee * Available
0.01
[0.90]
-0.15
[0.31]
0.06*
[0.08]
-0.05
[0.15]
Available
-0.47
[0.63]
-1.16
[0.62]
-0.12
[0.84]
-1.18
[0.18]
N
R-squared
52
0.18
52
0.14
52
0.08
52
0.19
35
© 2010
!
Some possible effects from the cross-section:
– Stocks with high expected loan fees
(high demand relative to supply of shorting)
» Experience more volatility in lending period
» Reverses in recall period
» Experience more negative skewness in lending period
» Reverses in recall period
– Consistent with a destabilizing effect for highest demand stocks
!
But results not very consistent and largely support earlier average
results that not much is affected
36
© 2010
Available stocks only (eliminating exogenous treatment)
First Phase
Dependent variable:
Second Phase
Return
Return
Panel B: Lending period, available stocks only
Short interest
0.03
[0.29]
-0.05***
[0.00]
0.01
[0.68]
0.00
[0.73]
0.00
[0.44]
-0.03
[0.34]
0.00
[0.87]
0.00
[0.40]
Potential loan to inst. own. (%)
-0.22
[0.25]
0.19
[0.22]
0.00
[1.00]
0.00
[0.84]
0.00
[0.50]
0.02*
[0.08]
0.06***
[0.00]
-0.00**
[0.05]
Expected loan fee
0.02
[0.66]
0.05
[0.12]
-0.03
[0.34]
0.00
[0.55]
0.00
[0.62]
-0.14***
[0.00]
-0.06*
[0.09]
-0.01
[0.13]
Constant
-0.22
[0.77]
0.31
[0.70]
-0.08
[0.92]
0.00
[0.98]
0.30**
[0.02]
-0.83
[0.11]
0.02
[0.96]
0.00
[0.95]
N
R-squared
32.00
0.25
32.00
0.18
32.00
0.04
32.00
0.02
19.00
0.03
19.00
0.35
19.00
0.30
19.00
0.11
37
© 2010
!
Larger number of significant results, particularly in second phase
– But relationships are spurious
– Without control group, might make erroneous inferences
!
Highlights importance of controlling for exogenous selection and the
experimental design
38
© 2010
Robustness
!
Other perturbations of the sample:
» Including all available and withheld stocks (40 vs. 20; 23 vs. 9)
» Removing the high loan fee stocks the Manager requested we
lend (37 vs. 20; 22 vs. 9)
» Removing all stocks whose loan fees decline to < 25 bps before
the lending program (32 vs. 17; 19 vs. 8)
» Removing the high loan fee stocks the Manager requested we
lend and stocks whose loan fees decline to < 25 bps before the
lending program (29 vs. 17; 18 vs. 8)
!
Results available in Appendix and results are unchanged.
39
© 2010
Implications
!
Theory:
– We find supply shocks impact loan fees, but not asset prices
– Why?
» Duffie, Garleanu, and Pedersen (2002) show direct link between
fees and prices, though link is not one-to-one
» Diamond and Verrechia (1987) show no price effect, but have no
loan fees in their model
– Perhaps loan fees are not market clearing (agents ration shares,
bundle them with other services)
– Or, market for stock borrowing segmented from underlying market
» Hedgers may exhibit price pressure in loan market but are too
small to affect underlying share prices
!
Future work to consider wedge between lending and underlying markets
40
© 2010
Implications
!
Fund manager behavior:
– *Initial motivation for the experiment
– Our results suggest managers (at the margin) can earn revenue from
lending shares without adversely affecting share prices
– Our Manager came to this conclusion, lifting restrictions on Oct. 1
– How much revenue from lending?
» Use all available + withheld stocks (total opportunity cost)
» Use potential loan amount (roughly 65-70% of ownership)
» Use three sets of fees: expected fees, actual fees + Data
Explorers, actual + hypothetical fees
» Can only calculate marginal revenue
– Difficult to generalize and can’t say much about equilibrium effects
» All else must be equal
41
© 2010
Table VIII: Marginal Revenue from Lending
!
Numbers represent returns % per year per dollar ownership in high
revenue stocks (which account for about 20% of $AUM)
Potential Fee Revenue per $ Owned of Stocks on Loan (in %)
!
First Phase
Mean
Median
Second Phase
Mean
Median
Expected loan fee
4.64
1.29
2.71
0.94
Actual loan fee
3.02
1.10
0.60
0.41
Actual (hypothetical) loan fee
2.78
0.83
0.49
0.35
First phase higher demand for shorting " higher fees
42
© 2010
Implications
!
Policy:
– Results suggest policies designed to restrict or alter supply of shares
will not be effective or useful.
» Consistent with other studies on the supply side. [Cohen, Diether,
and Malloy (2007), Diether, Lee, and Werner (2009)]
» We focus on very high demand stocks.
– Caution: Hard to assess general equilibrium effects.
– Policy debates could be better informed by experiments, where one
channel can be isolated holding everything else fixed.
43
© 2010
Conclusion
!
!
!
!
Experiment producing exogenous and sizeable supply shocks to
lendable shares
We find no adverse affects on underlying stock prices, despite moving
loan fees
Cross-sectional analysis yields similar findings
No evidence of supply effects for level of supply changes we generate
!
Implications for theory, manager behavior and policy
– Strong conclusions for money manager behavior at the margin
– Harder to draw conclusions for general equilibrium
– May inform policy debates
!
More experimentation worthwhile and useful
44
© 2010
Thank you.
Steve Kaplan
University of Chicago Booth School of Business and NBER
skaplan@uchicago.edu
45
© 2010
46
© 2010
47
© 2010
Table II: Summary Statistics
!
For days on which a security was on loan:
– Average loan fee = 2.6% (first phase), 1.6% (second phase)
» Varies from 1.4 to 976 bps
– Shares on loan = 5% (first phase), 9% (second phase) of short
interest
» Varies from 0.12% to 16%
– Average # shares lent 43% (first phase), 70% (second phase) of
daily trading volume (off-exchange volume not included)
– Average daily shares on loan about 1% of market cap
– Maximum daily loan about 1.6% of total market cap
» Varies from 0.03% to 5%
!
Increase to shares available for lending a non-trivial amount and lots of
cross-sectional variation
48
© 2010
Download